Z.22: When the Dashboard Beats Back
I wrote about some of the key characteristics of bad dashboards last fall.
But for many of the army’s junior commanders, the chromatic assaults on the senses aren’t the worst part. The real struggle lies in getting beaten over the head via the ugly dashboards built on months old data, usually by a commander who is a data toddler. Masked by too many superfluous pie charts, a lot of the army’s dashboards are laden with dated and irrelevant data. While PowerBI — and now GenAI — are available across the DoD, we still aren’t holding our commanders to any data literacy standard. Giving these leaders a custom-built dashboard is like giving your third grader the keys to a Ferrari.
The recent push to leverage data has seen a big step forward with Vantage. Run by Palantir, Vantage attempts to stitch together the hundreds of isolated data sources across the army, a daunting challenge to be sure. They’re making some slow progress and rolling out dashboards to showcase their results. Rumors have popped up that the Pentagon is shifting from PowerPoint to Vantage for USR reports. Effecting this culture change must be daunting, and I am wholeheartedly in favor of the drive to leverage data and kill PowerPoint.
But we’re not there yet. We’re held back not just by data literacy, but by shoddy data systems with no APIs.1 While Vantage can pull all these threads into a single — albeit cluttered — pane of glass, it cannot write back to these databases.
Laaaaaaaaaggggggg
When the data’s all siloed and there’s no APIs, you’re left with a dashboard littered with lagging data. Some of these lags are pretty egregious. MEDPROS, the system the Army uses to track our medical readiness, is perhaps the most egregious offender, often taking months to reflect updates. Which is absurd: In 1860 the Pony Express could be relied upon to take less than two weeks.
Across the army today, brigade and division commanders are beating up captains over Vantage dashboards that are showing months old data. These commanders are shooting well behind the duck, and don’t even realize it’s their own dashboards that’s causing the problem. And to be clear: when the dashboard sucks it’s the commander who is at fault.
In my combat deployments, we would often have drones stare at a target for days. This recce was the buy-in for lethal targeting when you’re doing everything you can to minimize unnecessary casualties.2 If our drone got pulled and we lost coverage we delayed our strike until we could regain a current sight picture of the objective.
This isn’t a unique rule to SOF. Even the conventional army wouldn’t choose to execute an operation off stale recce. No unit showing up to an NTC or JRTC rotation is going to conduct their attack based of the last rotation’s month-old recce. They’d all insist on getting their own recce out and getting a fresh picture of the battlefield. If they didn’t, we’d fire them.
Data is your recce
Too few commanders are asking questions of their dashboards. They aren’t asking how the data was gathered, how often it’s being refreshed, or even if it’s the right data in the first place. Stormking Analytics has been dropping a series on the challenges of dashboards.
Their posts do a great job of challenging leaders to understand what dashboards do, and do not do. Dashboards won’t ‘Decide What You Won’t’. They ‘can show you where you disagree, but they cannot tell you how to agree’. Stormking’s series illustrates how ‘dashboard factories’ aren’t going to magically make your organization data literate, as well as explaining why your subordinates are still using excel instead of your dashboard to do their jobs.
Enterprise systems are built to deliver accuracy, standardization, and consistency at scale. Operational decision-support tools must prioritize timeliness, adaptability, and proximity to action. These are different design objectives, and each serves a distinct purpose. The mistake is assuming that enterprise data can serve both equally well. When leaders expect enterprise platforms to handle every operational need, teams either slow down or quietly build around them. In most environments, they choose the latter.
Stormking’s series is well worth a read. They highlight the same problem: leaders must own their data. They need to be intentional in what they want it to do for them. Today, our worst problem is often IPPS-A’s laggy data — and shoddy Ux. This leads to wasted staff and soldier churn as everyone fights to ‘make the slide green’ in a system that’s always days, weeks, and months behind. That time would better be spent getting ready.
Because we need to be getting ready. In our next fight, commanders will have to fight against adversaries actively trying to corrupt their data. Our adversaries are training right now to execute the same deceptions operations that have been the hallmark of great military campaigns since antiquity. Only now they’re working to deceive our sensors and infect our data streams.
Data is your recce. And just like when you push out scouts forward, you need to fight for it.
Application Programming Interface. Interfaces that allow software to connect without a human in the loop. In this case, the ability to update the data without some person having to do it by hand.
